Journal article
Graph-based impact analysis as a framework for incorporating practitioner knowledge in dairy herd health management



Publication Details
Authors:
Krieger, M.; Schwabenbauer, E.; Hoischen-Taubner, S.; Emanuelson, U.; Sundrum, A.
Publication year:
2018
Journal:
Animal: An International Journal of Animal Bioscience
Pages range:
624-633
Journal acronym:
Animal
Volume number:
12
Issue number:
3
ISSN:
1751-7311
eISSN:
1751-732X
Languages:
English

Abstract

Production diseases in dairy cows are multifactorial, which means they
emerge from complex interactions between many different farm variables.
Variables with a large impact on production diseases can be identified
for groups of farms using statistical models, but these methods cannot
be used to identify highly influential variables in individual farms.
This, however, is necessary for herd health planning, because farm
conditions and associated health problems vary largely between farms.
The aim of this study was to rank variables according to their
anticipated effect on production diseases on the farm level by applying a
graph-based impact analysis on 192 European organic dairy farms. Direct
impacts between 13 pre-defined variables were estimated for each farm
during a round-table discussion attended by practitioners, that is
farmer, veterinarian and herd advisor. Indirect impacts were elaborated
through graph analysis taking into account impact strengths. Across
farms, factors supposedly exerting the most influence on production
diseases were ‘feeding’, ‘hygiene’ and ‘treatment’ (direct impacts), as
well as ‘knowledge and skills’ and ‘herd health monitoring’ (indirect
impacts). Factors strongly influenced by production diseases were ‘milk
performance’, ‘financial resources’ and ‘labour capacity’ (directly and
indirectly). Ranking of variables on the farm level revealed
considerable differences between farms in terms of their most
influential and most influenced farm factors. Consequently, very
different strategies may be required to reduce production diseases in
these farms. The method is based on perceptions and estimations and thus
prone to errors. From our point of view, however, this weakness is
clearly outweighed by the ability to assess and to analyse farm-specific
relationships and thus to complement general knowledge with contextual
knowledge. Therefore, we conclude that graph-based impact analysis
represents a promising decision support tool for herd health planning.
The next steps include testing the method using more specific and
problem-oriented variables as well as evaluating its effectiveness.


Last updated on 2019-01-08 at 14:32